Random Forest
Random Forest Prediction Model
Open libraries.
library("mlbench")
library("dplyr")
library("caret")
library("randomForest")
library("lattice")
library("ggplot2")
library("rpart")
library("e1071")
library("caret", lib.loc="/Library/Frameworks/R.framework/Versions/3.4/Resources/library")
library("stats")
library(relaimpo)
library(party)Get a list of Variable names.
Names <- names(dfNormDatAndFactor)
unlist(Names, recursive = TRUE, use.names = TRUE) [1] "Enrolling"
[2] "Sex.F"
[3] "Sex.M"
[4] "Expel.N"
[5] "Expel.Y"
[6] "First.Gen.N"
[7] "First.Gen.Y"
[8] "Challenge.Tag.N"
[9] "Challenge.Tag.Y"
[10] "Boettcher.Semi.N"
[11] "Boettcher.Semi.Y"
[12] "Boettcher.Final.N"
[13] "Boettcher.Final.Y"
[14] "Daniels.Semi.N"
[15] "Daniels.Semi.Y"
[16] "Daniels.Final.N"
[17] "Daniels.Final.Y"
[18] "Harvey.App.N"
[19] "Harvey.App.Y"
[20] "Harvey.Final.N"
[21] "Harvey.Final.Y"
[22] "FC.App.N"
[23] "FC.App.Y"
[24] "Thorson.App.N"
[25] "Thorson.App.Y"
[26] "Thorson.Admit.N"
[27] "Thorson.Admit.Y"
[28] "Summet.Participant.N"
[29] "Summet.Participant.Y"
[30] "Mines.Medal.N"
[31] "Mines.Medal.Y"
[32] "SPS.N"
[33] "SPS.Y"
[34] "Veteran.N"
[35] "Veteran.Y"
[36] "Legacy.N"
[37] "Legacy.Y"
[38] "Athlete.N"
[39] "Athlete.Y"
[40] "State.CO"
[41] "State.Other"
[42] "Citizenship.Foreign.National.International"
[43] "Citizenship.International"
[44] "Citizenship.Missing"
[45] "Citizenship.U.S..Citizen"
[46] "Citizenship.U.S..Permanent.Resident.Green.Card.Holder"
[47] "Citizenship.Undocumented.DACA"
[48] "Ethnicity.American.Indian.or.Alaska.Native"
[49] "Ethnicity.Asian"
[50] "Ethnicity.Black.or.African.American"
[51] "Ethnicity.Hispanic.or.Latino"
[52] "Ethnicity.Missing"
[53] "Ethnicity.Multiracial"
[54] "Ethnicity.Native.Hawaiian.or.Other.Pacific.Islander"
[55] "Ethnicity.NotDeclared"
[56] "Ethnicity.Unknown"
[57] "Ethnicity.White"
[58] "Major.App.Applied.Mathematics...Statistics...Computational...Applied.Mathematics"
[59] "Major.App.Applied.Mathematics...Statistics...Statistics"
[60] "Major.App.Chemical.Engineering"
[61] "Major.App.Chemical.Engineering...Biological.Engineering.Specialty"
[62] "Major.App.Chemistry"
[63] "Major.App.Chemistry...Biochemistry.Specialty"
[64] "Major.App.Chemistry...Environmental.Chemistry.Specialty"
[65] "Major.App.Civil.Engineering"
[66] "Major.App.Computer.Science"
[67] "Major.App.Economics"
[68] "Major.App.Electrical.Engineering"
[69] "Major.App.Engineering.Physics"
[70] "Major.App.Environmental.Engineering"
[71] "Major.App.Geological.Engineering"
[72] "Major.App.Geophysical.Engineering"
[73] "Major.App.Mechanical.Engineering"
[74] "Major.App.Metallurgical...Materials.Engineering"
[75] "Major.App.Mining.Engineering"
[76] "Major.App.Missing"
[77] "Major.App.Petroleum.Engineering"
[78] "Major.App.Undecided"
[79] "First.Contact.ACT"
[80] "First.Contact.Application"
[81] "First.Contact.Athlete.Form"
[82] "First.Contact.Campus.Visit"
[83] "First.Contact.College.Fair"
[84] "First.Contact.FUF"
[85] "First.Contact.Girls.Lead.the.Way"
[86] "First.Contact.GPA.Form"
[87] "First.Contact.Inquiry.Form"
[88] "First.Contact.Mailing"
[89] "First.Contact.Materials"
[90] "First.Contact.MEP"
[91] "First.Contact.Phone"
[92] "First.Contact.Preview.Mines"
[93] "First.Contact.Royall.Search"
[94] "First.Contact.SAT"
[95] "First.Contact.TOEFL"
[96] "First.Visit.Campus.Tour"
[97] "First.Visit.Campus.Visit"
[98] "First.Visit.Class.Shadow"
[99] "First.Visit.Discover.Mines"
[100] "First.Visit.Discovery.Mines.in.Your.City"
[101] "First.Visit.Girls.Lead.the.Way"
[102] "First.Visit.Launch"
[103] "First.Visit.Making.The.Connection"
[104] "First.Visit.Meet.Me.at.Mines"
[105] "First.Visit.None"
[106] "First.Visit.Preview.Mines"
[107] "First.Visit.Preview.Mines.In.Your.City"
[108] "Sport.Men.s.Varsity.Baseball"
[109] "Sport.Men.s.Varsity.Basketball"
[110] "Sport.Men.s.Varsity.Football"
[111] "Sport.Men.s.Varsity.Golf"
[112] "Sport.Men.s.Varsity.Soccer"
[113] "Sport.Men.s.Varsity.Wrestling"
[114] "Sport.None"
[115] "Sport.Varsity.Swimming.Diving"
[116] "Sport.Varsity.Track.Field"
[117] "Sport.Varsity.XCountry"
[118] "Sport.Women.s.Varsity.Basketball"
[119] "Sport.Women.s.Varsity.Soccer"
[120] "Sport.Women.s.Varsity.Softball"
[121] "Sport.Women.s.Varsity.Volleyball"
[122] "App.Created.Days"
[123] "Age"
[124] "HS.GPA"
[125] "SATR.Converted"
[126] "Review.OutsideActivity"
[127] "Review.Leadership"
[128] "Review.WorkExp"
[129] "Review.WorkEthic"
[130] "Review.ExpDiversity"
[131] "Review.DesireAttend"
[132] "Review.Affinity"
[133] "Review.InnovEntrep"
[134] "Review.Teamwork"
[135] "Review.OverallFit"
[136] "Logins.60Days"
[137] "EventCount.All"
[138] "EventCount.Admitted"
[139] "EventCount.Campus" Parrtitions created with 75% of data for training and 25% of data for testing.
inTraining <- createDataPartition(dfNormDatAndFactor$Enrolling, p = 0.75, list = FALSE)
training <- dfNormDatAndFactor[inTraining, ]
testing <- dfNormDatAndFactor[-inTraining, ]trainingtestingRun Random Forest Model with all variables.
rfModel <- randomForest(Enrolling ~ Sex.F + Sex.M +Expel.N + Expel.Y+ First.Gen.N+ First.Gen.Y+Challenge.Tag.N + Challenge.Tag.Y+ Boettcher.Semi.N+ Boettcher.Semi.Y+ Boettcher.Final.N+Boettcher.Final.Y + Daniels.Semi.N+Daniels.Semi.Y+ Daniels.Final.N+ Daniels.Final.Y+ Harvey.App.N+ Harvey.App.Y + Harvey.Final.N+ Harvey.Final.Y+ FC.App.N+ FC.App.Y+ Thorson.App.N+ Thorson.App.Y+ Thorson.Admit.N +Thorson.Admit.Y+Summet.Participant.N+ Summet.Participant.Y + Mines.Medal.N+ Mines.Medal.Y+ SPS.N+ SPS.Y+ Veteran.N+ Veteran.Y+ Legacy.N + Legacy.Y+ Athlete.N + Athlete.N + Athlete.Y + State.CO + State.Other + Citizenship.Foreign.National.International +Citizenship.International + Citizenship.Missing+ Citizenship.U.S..Citizen+ Citizenship.U.S..Permanent.Resident.Green.Card.Holder+ Citizenship.Undocumented.DACA+ Ethnicity.American.Indian.or.Alaska.Native+ Ethnicity.Asian+ Ethnicity.Black.or.African.American+ Ethnicity.Hispanic.or.Latino + Ethnicity.Missing + Ethnicity.Multiracial +Ethnicity.Native.Hawaiian.or.Other.Pacific.Islander + Ethnicity.NotDeclared + Ethnicity.Unknown +Ethnicity.White+ Major.App.Applied.Mathematics...Statistics...Computational...Applied.Mathematics+ Major.App.Applied.Mathematics...Statistics...Statistics+ Major.App.Chemical.Engineering+ Major.App.Chemical.Engineering...Biological.Engineering.Specialty+ Major.App.Chemistry+ Major.App.Chemistry...Biochemistry.Specialty + Major.App.Chemistry...Environmental.Chemistry.Specialty + Major.App.Civil.Engineering + Major.App.Computer.Science + Major.App.Economics+ Major.App.Electrical.Engineering+ Major.App.Engineering.Physics+ Major.App.Environmental.Engineering+ Major.App.Geological.Engineering + Major.App.Geophysical.Engineering +Major.App.Mechanical.Engineering + Major.App.Metallurgical...Materials.Engineering + Major.App.Mining.Engineering + Major.App.Missing + Major.App.Petroleum.Engineering+ Major.App.Undecided+ First.Contact.ACT + First.Contact.Application + First.Contact.Athlete.Form + First.Contact.Campus.Visit + First.Contact.College.Fair + First.Contact.FUF+First.Contact.Girls.Lead.the.Way+First.Contact.GPA.Form + First.Contact.Inquiry.Form+First.Contact.Mailing+ First.Contact.Materials + First.Contact.MEP+ First.Contact.Phone+ First.Contact.Preview.Mines+ First.Contact.Royall.Search+First.Contact.TOEFL+ First.Contact.SAT+ First.Visit.Campus.Tour+ First.Visit.Campus.Visit+ First.Visit.Class.Shadow + First.Visit.Discover.Mines+ First.Visit.Discovery.Mines.in.Your.City+ First.Visit.Girls.Lead.the.Way+ First.Visit.Launch + First.Visit.Making.The.Connection + First.Visit.Meet.Me.at.Mines+ First.Visit.None+ First.Visit.Preview.Mines + First.Visit.Preview.Mines.In.Your.City + Sport.Men.s.Varsity.Baseball+ Sport.Men.s.Varsity.Basketball+ Sport.Men.s.Varsity.Football + Sport.Men.s.Varsity.Golf +Sport.Men.s.Varsity.Soccer+ Sport.Men.s.Varsity.Wrestling+ Sport.None+ Sport.Varsity.Swimming.Diving+ Sport.Varsity.Track.Field+Sport.Varsity.XCountry+ Sport.Women.s.Varsity.Basketball + Sport.Women.s.Varsity.Soccer+ Sport.Women.s.Varsity.Softball + Sport.Women.s.Varsity.Volleyball+ App.Created.Days + Age + HS.GPA+ SATR.Converted +Review.OutsideActivity +Review.Leadership+Review.WorkExp+ Review.WorkEthic +Review.ExpDiversity +Review.DesireAttend +Review.Affinity+ Review.InnovEntrep+Review.Teamwork +Review.OverallFit + Logins.60Days+EventCount.All+EventCount.Admitted+EventCount.Campus, data = training)
rfModel
Call:
randomForest(formula = Enrolling ~ Sex.F + Sex.M + Expel.N + Expel.Y + First.Gen.N + First.Gen.Y + Challenge.Tag.N + Challenge.Tag.Y + Boettcher.Semi.N + Boettcher.Semi.Y + Boettcher.Final.N + Boettcher.Final.Y + Daniels.Semi.N + Daniels.Semi.Y + Daniels.Final.N + Daniels.Final.Y + Harvey.App.N + Harvey.App.Y + Harvey.Final.N + Harvey.Final.Y + FC.App.N + FC.App.Y + Thorson.App.N + Thorson.App.Y + Thorson.Admit.N + Thorson.Admit.Y + Summet.Participant.N + Summet.Participant.Y + Mines.Medal.N + Mines.Medal.Y + SPS.N + SPS.Y + Veteran.N + Veteran.Y + Legacy.N + Legacy.Y + Athlete.N + Athlete.N + Athlete.Y + State.CO + State.Other + Citizenship.Foreign.National.International + Citizenship.International + Citizenship.Missing + Citizenship.U.S..Citizen + Citizenship.U.S..Permanent.Resident.Green.Card.Holder + Citizenship.Undocumented.DACA + Ethnicity.American.Indian.or.Alaska.Native + Ethnicity.Asian + Ethnicity.Black.or.African.American + Ethnicity.Hispanic.or.Latino + Ethnicity.Missing + Ethnicity.Multiracial + Ethnicity.Native.Hawaiian.or.Other.Pacific.Islander + Ethnicity.NotDeclared + Ethnicity.Unknown + Ethnicity.White + Major.App.Applied.Mathematics...Statistics...Computational...Applied.Mathematics + Major.App.Applied.Mathematics...Statistics...Statistics + Major.App.Chemical.Engineering + Major.App.Chemical.Engineering...Biological.Engineering.Specialty + Major.App.Chemistry + Major.App.Chemistry...Biochemistry.Specialty + Major.App.Chemistry...Environmental.Chemistry.Specialty + Major.App.Civil.Engineering + Major.App.Computer.Science + Major.App.Economics + Major.App.Electrical.Engineering + Major.App.Engineering.Physics + Major.App.Environmental.Engineering + Major.App.Geological.Engineering + Major.App.Geophysical.Engineering + Major.App.Mechanical.Engineering + Major.App.Metallurgical...Materials.Engineering + Major.App.Mining.Engineering + Major.App.Missing + Major.App.Petroleum.Engineering + Major.App.Undecided + First.Contact.ACT + First.Contact.Application + First.Contact.Athlete.Form + First.Contact.Campus.Visit + First.Contact.College.Fair + First.Contact.FUF + First.Contact.Girls.Lead.the.Way + First.Contact.GPA.Form + First.Contact.Inquiry.Form + First.Contact.Mailing + First.Contact.Materials + First.Contact.MEP + First.Contact.Phone + First.Contact.Preview.Mines + First.Contact.Royall.Search + First.Contact.TOEFL + First.Contact.SAT + First.Visit.Campus.Tour + First.Visit.Campus.Visit + First.Visit.Class.Shadow + First.Visit.Discover.Mines + First.Visit.Discovery.Mines.in.Your.City + First.Visit.Girls.Lead.the.Way + First.Visit.Launch + First.Visit.Making.The.Connection + First.Visit.Meet.Me.at.Mines + First.Visit.None + First.Visit.Preview.Mines + First.Visit.Preview.Mines.In.Your.City + Sport.Men.s.Varsity.Baseball + Sport.Men.s.Varsity.Basketball + Sport.Men.s.Varsity.Football + Sport.Men.s.Varsity.Golf + Sport.Men.s.Varsity.Soccer + Sport.Men.s.Varsity.Wrestling + Sport.None + Sport.Varsity.Swimming.Diving + Sport.Varsity.Track.Field + Sport.Varsity.XCountry + Sport.Women.s.Varsity.Basketball + Sport.Women.s.Varsity.Soccer + Sport.Women.s.Varsity.Softball + Sport.Women.s.Varsity.Volleyball + App.Created.Days + Age + HS.GPA + SATR.Converted + Review.OutsideActivity + Review.Leadership + Review.WorkExp + Review.WorkEthic + Review.ExpDiversity + Review.DesireAttend + Review.Affinity + Review.InnovEntrep + Review.Teamwork + Review.OverallFit + Logins.60Days + EventCount.All + EventCount.Admitted + EventCount.Campus, data = training)
Type of random forest: classification
Number of trees: 500
No. of variables tried at each split: 11
OOB estimate of error rate: 8.62%
Confusion matrix:
N Y class.error
N 3563 127 0.03441734
Y 276 711 0.27963526rfModel.prediction <- predict(rfModel, testing)
table(rfModel.prediction, testing$Enrolling)
rfModel.prediction N Y
N 1181 96
Y 48 233To determine variable importance.
imprfModel <- importance(rfModel)
imprfModel MeanDecreaseGini
Sex.F 7.437706e+00
Sex.M 7.970186e+00
Expel.N 8.659331e-01
Expel.Y 9.745556e-01
First.Gen.N 5.083216e+00
First.Gen.Y 5.162621e+00
Challenge.Tag.N 2.142498e+00
Challenge.Tag.Y 2.090986e+00
Boettcher.Semi.N 1.268805e+00
Boettcher.Semi.Y 1.088516e+00
Boettcher.Final.N 4.735714e-01
Boettcher.Final.Y 4.705432e-01
Daniels.Semi.N 1.967173e+00
Daniels.Semi.Y 1.647909e+00
Daniels.Final.N 4.607632e-01
Daniels.Final.Y 3.839780e-01
Harvey.App.N 4.454943e+00
Harvey.App.Y 4.307206e+00
Harvey.Final.N 1.489947e+00
Harvey.Final.Y 1.734864e+00
FC.App.N 2.932341e+00
FC.App.Y 3.056475e+00
Thorson.App.N 6.108025e+00
Thorson.App.Y 5.748607e+00
Thorson.Admit.N 5.450054e+00
Thorson.Admit.Y 5.647499e+00
Summet.Participant.N 8.726508e-01
Summet.Participant.Y 8.859990e-01
Mines.Medal.N 9.469697e-05
Mines.Medal.Y 3.118012e-04
SPS.N 1.104065e+00
SPS.Y 1.168917e+00
Veteran.N 3.948454e+00
Veteran.Y 3.933734e+00
Legacy.N 6.212300e+00
Legacy.Y 6.580279e+00
Athlete.N 6.485726e+00
Athlete.Y 6.714595e+00
State.CO 3.247304e+01
State.Other 2.920199e+01
Citizenship.Foreign.National.International 3.876040e+00
Citizenship.International 3.288128e-01
Citizenship.Missing 2.853720e+00
Citizenship.U.S..Citizen 5.462649e+00
Citizenship.U.S..Permanent.Resident.Green.Card.Holder 1.888712e+00
Citizenship.Undocumented.DACA 1.307468e-01
Ethnicity.American.Indian.or.Alaska.Native 1.831798e+00
Ethnicity.Asian 4.660927e+00
Ethnicity.Black.or.African.American 1.244916e+00
Ethnicity.Hispanic.or.Latino 3.141938e+00
Ethnicity.Missing 4.584504e-01
Ethnicity.Multiracial 0.000000e+00
Ethnicity.Native.Hawaiian.or.Other.Pacific.Islander 1.931345e-01
Ethnicity.NotDeclared 3.649621e+00
Ethnicity.Unknown 0.000000e+00
Ethnicity.White 7.932935e+00
Major.App.Applied.Mathematics...Statistics...Computational...Applied.Mathematics 1.583910e+00
Major.App.Applied.Mathematics...Statistics...Statistics 8.705972e-01
Major.App.Chemical.Engineering 7.347701e+00
Major.App.Chemical.Engineering...Biological.Engineering.Specialty 5.192139e+00
Major.App.Chemistry 1.054614e+00
Major.App.Chemistry...Biochemistry.Specialty 1.475858e+00
Major.App.Chemistry...Environmental.Chemistry.Specialty 2.618606e-01
Major.App.Civil.Engineering 4.720395e+00
Major.App.Computer.Science 4.695704e+00
Major.App.Economics 1.010823e+00
Major.App.Electrical.Engineering 3.748680e+00
Major.App.Engineering.Physics 5.395229e+00
Major.App.Environmental.Engineering 3.140402e+00
Major.App.Geological.Engineering 2.851079e+00
Major.App.Geophysical.Engineering 7.893425e-01
Major.App.Mechanical.Engineering 8.485352e+00
Major.App.Metallurgical...Materials.Engineering 2.134452e+00
Major.App.Mining.Engineering 2.443291e+00
Major.App.Missing 2.788781e+00
Major.App.Petroleum.Engineering 4.397457e+00
Major.App.Undecided 5.302115e+00
First.Contact.ACT 6.692925e+00
First.Contact.Application 8.421762e+00
First.Contact.Athlete.Form 1.114967e+00
First.Contact.Campus.Visit 1.347537e+01
First.Contact.College.Fair 2.625882e-01
First.Contact.FUF 7.978287e+00
First.Contact.Girls.Lead.the.Way 5.371549e-01
First.Contact.GPA.Form 8.735425e-01
First.Contact.Inquiry.Form 2.369550e+00
First.Contact.Mailing 5.415302e-01
First.Contact.Materials 3.451884e-01
First.Contact.MEP 8.260278e-01
First.Contact.Phone 2.024735e-02
First.Contact.Preview.Mines 1.464733e+00
First.Contact.Royall.Search 1.072953e+01
First.Contact.TOEFL 7.034645e-01
First.Contact.SAT 1.420114e+00
First.Visit.Campus.Tour 1.370637e+01
First.Visit.Campus.Visit 1.480266e+01
First.Visit.Class.Shadow 2.329774e+00
First.Visit.Discover.Mines 7.929069e+00
First.Visit.Discovery.Mines.in.Your.City 1.363295e+00
First.Visit.Girls.Lead.the.Way 5.880593e-01
First.Visit.Launch 4.658352e+01
First.Visit.Making.The.Connection 4.584449e+00
First.Visit.Meet.Me.at.Mines 1.263360e+00
First.Visit.None 6.927932e+01
First.Visit.Preview.Mines 6.506139e+00
First.Visit.Preview.Mines.In.Your.City 1.782319e+01
Sport.Men.s.Varsity.Baseball 1.934009e-01
Sport.Men.s.Varsity.Basketball 4.905744e-01
Sport.Men.s.Varsity.Football 2.877780e+00
Sport.Men.s.Varsity.Golf 2.163090e-01
Sport.Men.s.Varsity.Soccer 4.367005e-01
Sport.Men.s.Varsity.Wrestling 6.442545e-01
Sport.None 6.618847e+00
Sport.Varsity.Swimming.Diving 1.314591e-02
Sport.Varsity.Track.Field 2.996113e-01
Sport.Varsity.XCountry 9.709506e-02
Sport.Women.s.Varsity.Basketball 4.858711e-01
Sport.Women.s.Varsity.Soccer 3.653904e-01
Sport.Women.s.Varsity.Softball 1.199113e-02
Sport.Women.s.Varsity.Volleyball 5.702375e-01
App.Created.Days 5.761980e+01
Age 4.282867e+01
HS.GPA 6.344325e+01
SATR.Converted 4.569617e+01
Review.OutsideActivity 1.585565e+01
Review.Leadership 1.912368e+01
Review.WorkExp 1.929378e+01
Review.WorkEthic 1.537185e+01
Review.ExpDiversity 1.606620e+01
Review.DesireAttend 2.316809e+01
Review.Affinity 1.992175e+01
Review.InnovEntrep 1.495064e+01
Review.Teamwork 1.910074e+01
Review.OverallFit 2.497267e+01
Logins.60Days 1.379456e+00
EventCount.All 1.642425e+02
EventCount.Admitted 2.249928e+02
EventCount.Campus 1.435702e+02#format((sort(imprfModel, decreasing=TRUE)), scientific=F)
#sort(imprfModel, decreasing=TRUE) # relative importancedfimprfModel <-as.data.frame(imprfModel)
dfimprfModelconfusionMatrix(table(rfModel.prediction, testing$Enrolling))Confusion Matrix and Statistics
rfModel.prediction N Y
N 1181 96
Y 48 233
Accuracy : 0.9076
95% CI : (0.8921, 0.9215)
No Information Rate : 0.7888
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 0.7069
Mcnemar's Test P-Value : 8.978e-05
Sensitivity : 0.9609
Specificity : 0.7082
Pos Pred Value : 0.9248
Neg Pred Value : 0.8292
Prevalence : 0.7888
Detection Rate : 0.7580
Detection Prevalence : 0.8196
Balanced Accuracy : 0.8346
'Positive' Class : N
Plot of importance of Variabels.
varImpPlot(rfModel,cex = .6, pt.cex = .7, color = "navy blue", gcolor = par("fg"), lcolor = "gray", main = "Variable Importance for Random Forest Model")
Model with less variables
rfModelLess <- randomForest(Enrolling ~ State.CO + State.Other + Major.App.Mechanical.Engineering + First.Contact.Application + First.Contact.Campus.Visit + First.Contact.FUF + First.Contact.Royall.Search+ First.Visit.Campus.Tour+ First.Visit.Campus.Visit+ First.Visit.Discover.Mines+ First.Visit.Launch + First.Visit.None+ First.Visit.Preview.Mines.In.Your.City + App.Created.Days + Age + HS.GPA+ SATR.Converted +Review.OutsideActivity +Review.Leadership+Review.WorkExp+ Review.WorkEthic +Review.ExpDiversity +Review.DesireAttend +Review.Affinity+ Review.InnovEntrep+Review.Teamwork +Review.OverallFit +Ethnicity.White +EventCount.All+EventCount.Admitted+EventCount.Campus, data = training)
rfModelLess
Call:
randomForest(formula = Enrolling ~ State.CO + State.Other + Major.App.Mechanical.Engineering + First.Contact.Application + First.Contact.Campus.Visit + First.Contact.FUF + First.Contact.Royall.Search + First.Visit.Campus.Tour + First.Visit.Campus.Visit + First.Visit.Discover.Mines + First.Visit.Launch + First.Visit.None + First.Visit.Preview.Mines.In.Your.City + App.Created.Days + Age + HS.GPA + SATR.Converted + Review.OutsideActivity + Review.Leadership + Review.WorkExp + Review.WorkEthic + Review.ExpDiversity + Review.DesireAttend + Review.Affinity + Review.InnovEntrep + Review.Teamwork + Review.OverallFit + Ethnicity.White + EventCount.All + EventCount.Admitted + EventCount.Campus, data = training)
Type of random forest: classification
Number of trees: 500
No. of variables tried at each split: 5
OOB estimate of error rate: 8.51%
Confusion matrix:
N Y class.error
N 3563 127 0.03441734
Y 271 716 0.27456940rfModelLess.prediction <- predict(rfModelLess, testing)
table(rfModelLess.prediction, testing$Enrolling)
rfModelLess.prediction N Y
N 1183 98
Y 46 231To determine variable imortance.
imprfModelLess <- importance(rfModelLess)
#format(imprfModelLess, scientific=F)
imprfModelLess MeanDecreaseGini
State.CO 30.878949
State.Other 29.257083
Major.App.Mechanical.Engineering 12.579230
First.Contact.Application 11.785755
First.Contact.Campus.Visit 15.618481
First.Contact.FUF 9.576268
First.Contact.Royall.Search 15.563429
First.Visit.Campus.Tour 15.820761
First.Visit.Campus.Visit 19.779964
First.Visit.Discover.Mines 9.325906
First.Visit.Launch 55.597020
First.Visit.None 82.567953
First.Visit.Preview.Mines.In.Your.City 16.833940
App.Created.Days 85.972986
Age 65.084346
HS.GPA 98.868442
SATR.Converted 72.984411
Review.OutsideActivity 21.638767
Review.Leadership 29.026726
Review.WorkExp 28.769419
Review.WorkEthic 22.232976
Review.ExpDiversity 22.870812
Review.DesireAttend 32.226286
Review.Affinity 29.324082
Review.InnovEntrep 23.136576
Review.Teamwork 28.812416
Review.OverallFit 37.286821
Ethnicity.White 14.279990
EventCount.All 193.936731
EventCount.Admitted 238.199556
EventCount.Campus 157.806502confusionMatrix(table(rfModelLess.prediction, testing$Enrolling))Confusion Matrix and Statistics
rfModelLess.prediction N Y
N 1183 98
Y 46 231
Accuracy : 0.9076
95% CI : (0.8921, 0.9215)
No Information Rate : 0.7888
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 0.7055
Mcnemar's Test P-Value : 2.138e-05
Sensitivity : 0.9626
Specificity : 0.7021
Pos Pred Value : 0.9235
Neg Pred Value : 0.8339
Prevalence : 0.7888
Detection Rate : 0.7593
Detection Prevalence : 0.8222
Balanced Accuracy : 0.8323
'Positive' Class : N
Model with top 25 variables
rfModelTop25 <- randomForest(Enrolling ~ State.CO + State.Other + First.Contact.Campus.Visit + First.Visit.Campus.Tour+ First.Visit.Campus.Visit+ First.Visit.Discover.Mines+ First.Visit.Launch + First.Visit.None+ First.Visit.Preview.Mines.In.Your.City + App.Created.Days + Age + HS.GPA+ SATR.Converted +Review.OutsideActivity +Review.Leadership+Review.WorkExp+ Review.WorkEthic +Review.ExpDiversity +Review.DesireAttend +Review.Affinity+ Review.InnovEntrep+Review.Teamwork +Review.OverallFit +EventCount.All+EventCount.Admitted+EventCount.Campus, data = training)
rfModelTop25
Call:
randomForest(formula = Enrolling ~ State.CO + State.Other + First.Contact.Campus.Visit + First.Visit.Campus.Tour + First.Visit.Campus.Visit + First.Visit.Discover.Mines + First.Visit.Launch + First.Visit.None + First.Visit.Preview.Mines.In.Your.City + App.Created.Days + Age + HS.GPA + SATR.Converted + Review.OutsideActivity + Review.Leadership + Review.WorkExp + Review.WorkEthic + Review.ExpDiversity + Review.DesireAttend + Review.Affinity + Review.InnovEntrep + Review.Teamwork + Review.OverallFit + EventCount.All + EventCount.Admitted + EventCount.Campus, data = training)
Type of random forest: classification
Number of trees: 500
No. of variables tried at each split: 5
OOB estimate of error rate: 8.66%
Confusion matrix:
N Y class.error
N 3565 125 0.03387534
Y 280 707 0.28368794rfModelTop25.prediction <- predict(rfModelTop25, testing)
table(rfModelTop25.prediction, testing$Enrolling)
rfModelTop25.prediction N Y
N 1182 98
Y 47 231To determine variable imortance.
imprfModelTop25 <- importance(rfModelTop25)
#format(imprfModelLess, scientific=F)
imprfModelTop25 MeanDecreaseGini
State.CO 32.81859
State.Other 29.26938
First.Contact.Campus.Visit 16.71002
First.Visit.Campus.Tour 16.26760
First.Visit.Campus.Visit 21.30551
First.Visit.Discover.Mines 10.33880
First.Visit.Launch 53.72968
First.Visit.None 76.59342
First.Visit.Preview.Mines.In.Your.City 17.63941
App.Created.Days 97.41080
Age 70.79303
HS.GPA 108.79595
SATR.Converted 79.43763
Review.OutsideActivity 22.15891
Review.Leadership 30.39901
Review.WorkExp 31.24720
Review.WorkEthic 23.82738
Review.ExpDiversity 23.54627
Review.DesireAttend 32.95253
Review.Affinity 30.80140
Review.InnovEntrep 24.81996
Review.Teamwork 30.19490
Review.OverallFit 39.41873
EventCount.All 198.86630
EventCount.Admitted 252.81695
EventCount.Campus 157.31704confusionMatrix(table(rfModelTop25.prediction, testing$Enrolling))Confusion Matrix and Statistics
rfModelTop25.prediction N Y
N 1182 98
Y 47 231
Accuracy : 0.9069
95% CI : (0.8914, 0.9209)
No Information Rate : 0.7888
P-Value [Acc > NIR] : < 2.2e-16
Kappa : 0.7038
Mcnemar's Test P-Value : 3.292e-05
Sensitivity : 0.9618
Specificity : 0.7021
Pos Pred Value : 0.9234
Neg Pred Value : 0.8309
Prevalence : 0.7888
Detection Rate : 0.7587
Detection Prevalence : 0.8216
Balanced Accuracy : 0.8319
'Positive' Class : N
PredictionTop25 <- cbind(rfModelTop25.prediction,testing)
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